Solid-state NMR structural measurements on the membrane-associated influenza fusion protein ectodomain.

Enveloped viruses such as HIV and influenza virus (IFV) are enclosed by a membrane which is obtained from an infected host cell. Infection of a new cell begins with joining or “fusion” of the viral and host cell membranes with an end result of a single membrane and the viral nucleocapsid in the host cell cytoplasm. Although membrane fusion is thermodynamically allowed, the rates of uncatalyzed membrane fusion are typically small. For this reason, enveloped viruses have fusion proteins in their membranes that bind to the host cell membranes and catalyze fusion.1 This paper describes studies of a large and functional domain of the IFV hemagglutinin (HA) fusion protein and includes bacterial expression and isotopic labeling of the protein, characterization of its folding and fusion activity, and membrane incorporation and solid-state nuclear magnetic resonance (SSNMR) structural measurements. The HA protein is composed of HA1 and HA2 subunits. HA1 lies completely outside the virus, while HA2 has a ∼185 residue N-terminal ectodomain that lies outside the virus, a ∼25 residue transmembrane domain, and a ∼10 residue C-terminal endodomain that is inside the virus.2 The IFV is taken into the host respiratory epithelial cell by receptor-mediated endocytosis, and the cell physiological processes lower the pH of the endosome to ∼5. The HA1 and HA2 subunits dissociate, and a large HA2 structural change results in exposure of the ∼20 residue N-terminal “fusion peptide” (IFP) region. The IFP binds to endosomal membranes, and membrane fusion occurs. There has been a pH 7.5 structure of the HA1/HA2 ectodomain complex crystallized from aqueous solution and a pH 4.4 structure of residues 34-178 of HA2 that forms the “soluble ectodomain” (SHA2) and which was also crystallized from aqueous solution.2,3 In addition, there have been liquid-state NMR structures of IFP in detergent micelles as well as electron spin resonance measurements of motion and membrane insertion of specific residues of IFP and of a HA2 construct composed of residues 1-127.4,5 The present work is on a “FHA2” full ectodomain construct composed of residues 1-185 of HA2 and an eight residue C-terminal tag (Figure 1A). SSNMR has the potential for providing high-resolution structural information for FHA2 in the physiologically relevant membrane-bound state and for addressing structural effects of factors that reduce fusion activity including neutral pH and mutations. There have been some previous applications of SSNMR to other large bacterial and human membrane proteins as well as membraneassociated IFP, and our study builds on this work.6-10 SSNMR requires efficient production of >10 mg quantities of isotopically labeled protein, and this was accomplished by FHA2 expression in Escherichia coli cells. Significant isotopic labeling requires expression in minimal media which lacks amino acids, but it was found that the purified FHA2 yield was ∼0.1 mg/L fermentation culture for E. coli grown only in minimal media. The successful approach was growth to OD 7 in a rich LB medium followed by a switch to minimal medium composed of glucose, salts, and the labeled amino acids.11 Although there has been progress in SSNMR assignment and structure determination of uniformly 13C,15N-labeled membrane proteins, it was decided to begin with amino acid type labeling so that assignment would be more straightforward.12-14 FHA2 purification was done using 0.5% N-laurylsarcosine detergent, and FHA2 with >95% purity was obtained using a cobalt resin which bound the FHA2 histidine tag (Figure 1B). Yields of ∼8 g cell mass and ∼3 mg purified FHA2 per liter fermentation culture were obtained with this approach. The FHA2 was exchanged into a solution of 0.5% â-octylglucoside detergent (BOG) in 5 mM HEPES/10 mM MES (“HM buffer”) at pH 7.4. The overall secondary structure as a function of pH was probed with circular dichroism (CD) spectroscopy (Figure 1C). Observation of CD minima at 208 and 222 nm at both pH 5.0 and pH 7.4 was consistent with a significant fraction of helical conformation. The θ222nm value of -16 000 deg‚cm2/dmol at pH 5.0 correlated with ∼50% of the residues in helical conformation and can be compared to the ∼60% of the residues in helical conformation expected if the SHA2 and IFP regions of detergent-associated FHA2 have the same conformations observed in their respective structures. A common assay to probe fusion peptide-induced membrane perturbation is peptide-induced lipid mixing (LM) between different unilamellar vesicles. By this assay, FHA2 was a potent fusogen and worked at ∼10-fold lower concentrations than has been observed for IFP (Figure 1D).15,16 At FHA2/lipid ∼ 0.001, the LM rate was >0.5 s-1 and is >10-fold larger than LM rates observed for fusion peptides at higher ratios.15 There was also a striking pH dependence of FHA2-induced lipid † Department of Chemistry. ‡ Department of Chemical Engineering. Figure 1. (A) FHA2 amino acid sequence from the influenza X31 strain. Each of the underlined residues is a first residue in a unique sequential pair. (B) SDS-PAGE gel of purified FHA2, MW ) 22.5 kD. (C) Circular dichroism spectra at 4 °C of FHA2 in 0.5% BOG detergent at pH 5.0 (red line) and pH 7.4 (black line). (D) Final extent of lipid mixing in vesicles of LM37 induced by FHA2 at pH 5.0 (filled bars) and pH 7.4 (open bars). Published on Web 08/25/2007